Date

5-1-2018

Document Type

Thesis

Degree

Master of Science

Department

Industrial Engineering

First Adviser

Defourny, Boris

Abstract

Wind ramp events have a significant influence of uncertainty in wind power production. In order to build an efficient decision-making systems for the smart grid, developing statistical models based on analysis of historical data of wind ramp events is indispensable. In this paper, we design a detection algorithm to analyze historical data, build distribution models to predict and simulate wind ramp events. Phase-type distribution consists of a convolution of the Exponential distribution which can be used to apply Markov decisions process and identify the factors which can cause wind ramp events. We use three types of Phase-type distribution to fit the data sets of duration, obtain the optimal number of phases and the parameters. Both the model of simulation and Phase-type distribution can be used to help making decisions and improving the accuracy of forecast for wind power production in smart grid.

Share

COinS